Background: Migraine is a common episodic brain disorder. Treatment options and diagnosis are hampered by an incomplete understanding of disease pathophysiology and the lack of objective diagnostic markers. The aim of this study was to identify biochemical differences characteristic for different subtypes of migraine in cerebrospinal fluid (CSF) of migraine patients using an exploratory 1H-NMR-based metabolomics approach.
Methods: CSF was obtained, in between migraine attacks, via lumbar puncture from patients with hemiplegic migraine, migraine with aura, migraine without aura, and healthy controls. Metabolite concentrations were measured by quantitative 1H-NMR spectroscopy. Multivariate data analysis was used to find the optimal set of predictors, generalized linear models (GLM) were used to ascertain the differential significance of individual metabolites.
Results: In CSF samples from 18 patients with hemiplegic migraine, 38 with migraine with aura, 27 migraine without aura, and 43 healthy controls, nineteen metabolites were identified and quantified. Hemiplegic migraine patients could be discriminated from healthy controls using supervised multivariate modelling with 2-hydroxybutyrate and 2-hydroxyisovalerate as the most discriminant metabolites. Univariate GLM analysis showed 2-hydroxybutyrate to be lower in hemiplegic migraine compared with healthy controls; no significant differences were observed for other metabolites. It was not possible to discriminate migraine with and without aura from healthy controls based on their metabolic profile.
Conclusions: Using an exploratory 1H-NMR metabolomics analysis we identified metabolites that were able to discriminate hemiplegic migraine patients from healthy controls. The lower levels of 2-hydroxybutyrate found in patients with hemiplegic migraine could indicate a dysregulation of the brain's energy metabolism. An experimental confirmation in vitro or in animal models will be required to confirm or discard this hypothesis. Migraine with and migraine without aura patients did not reveal a metabolic profile different from healthy controls.